ENCOW16BX (Schäfer & Bildhauer (2012))
webcorpus containing 9 billion tokens
downloaded data queried via CWB
[lemma=“let”] [pos=“N.|PP.”][pos=“V.*”][pos=“DT”]?[pos=“N.*”]
recurrent quotes, e.g. Let them eat cake
discourse-organizing and text-organizing semi-formulaic patterns, e.g. Let me explain, let us examine
proverbs, e.g. let bygones be bygones
Multiple covarying collexeme analysis (Stefanowitsch & Gries (2005))
two variants:
(pro)noun1 - verb - noun2
verb - noun2
importance of discourse-structuring constructions, both with communication verbs (Let me say something, let me ask/answer a/the question) and in more “text-structuring” contexts: let us take/have a look, let you know the results
fixed chunks repeated numerous times (eat - cake, bygones - bygones)
overall, strong dominance of “communicative” / communication-structuring contexts
complex NPs (more than one word) in the open slots disregarded
pronouns in the second open slot disregarded
some false positives due to annotation errors of the original data
context not taken into account (easy follow-up e.g. looking which items are preceded by don’t)